mcp-agents 0.12.1 → 0.12.2

package/README.md

@@ -88,7 +88,9 @@ or Gemini during bridge calls.
 
 Other startup defaults: `sandbox_mode=workspace-write`, `approval_policy=never`
 (both configurable via `--sandbox_mode` / `--approval_policy`, and steerable per
-call); `features.multi_agent=false` is fixed.
+call); `features.multi_agent=false`, `features.apps=false`, and
+`features.plugins=false` are fixed (the latter two keep ChatGPT app/plugin
+skills — Figma, Gmail, Presentations, etc. — out of the bridged session context).
 
 Startup flags (`--model`, `--model_reasoning_effort`) set the model and effort for the native Codex MCP server. Per-call `model` and the model/effort keys inside a `config` override are stripped from `tools/call` before they reach Codex, so a client cannot override the pinned model/effort for a single call (`sandbox`, `cwd`, and `approval-policy` — top-level and the matching `config` keys — are intentionally left steerable per call). For example, this request:
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mcp-agents",
-  "version": "0.12.1",
+  "version": "0.12.2",
   "description": "MCP server that wraps AI CLI tools (Claude Code, Gemini CLI, Codex CLI) for use by any MCP client",
   "type": "module",
   "bin": {

package/server.js

@@ -480,6 +480,8 @@ function buildCodexBridgeConfig({
     `sandbox_mode = ${toTomlString(sandboxMode)}`,
     "",
     "[features]",
+    "apps = false",
+    "plugins = false",
     "multi_agent = false",
     "",
   ].join("\n");